1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
|
;******************************************************************************
;* MMX optimized DSP utils
;* Copyright (c) 2008 Loren Merritt
;*
;* This file is part of FFmpeg.
;*
;* FFmpeg is free software; you can redistribute it and/or
;* modify it under the terms of the GNU Lesser General Public
;* License as published by the Free Software Foundation; either
;* version 2.1 of the License, or (at your option) any later version.
;*
;* FFmpeg is distributed in the hope that it will be useful,
;* but WITHOUT ANY WARRANTY; without even the implied warranty of
;* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
;* Lesser General Public License for more details.
;*
;* You should have received a copy of the GNU Lesser General Public
;* License along with FFmpeg; if not, write to the Free Software
;* 51, Inc., Foundation Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
;******************************************************************************
%include "x86inc.asm"
SECTION_RODATA
pb_f: times 16 db 15
pb_zzzzzzzz77777777: times 8 db -1
pb_7: times 8 db 7
pb_zzzz3333zzzzbbbb: db -1,-1,-1,-1,3,3,3,3,-1,-1,-1,-1,11,11,11,11
pb_zz11zz55zz99zzdd: db -1,-1,1,1,-1,-1,5,5,-1,-1,9,9,-1,-1,13,13
section .text align=16
%macro PSWAPD_SSE 2
pshufw %1, %2, 0x4e
%endmacro
%macro PSWAPD_3DN1 2
movq %1, %2
psrlq %1, 32
punpckldq %1, %2
%endmacro
%macro FLOAT_TO_INT16_INTERLEAVE6 1
; void ff_float_to_int16_interleave6_sse(int16_t *dst, const float **src, int len)
cglobal float_to_int16_interleave6_%1, 2,7,0, dst, src, src1, src2, src3, src4, src5
%ifdef ARCH_X86_64
%define lend r10d
mov lend, r2d
%else
%define lend dword r2m
%endif
mov src1q, [srcq+1*gprsize]
mov src2q, [srcq+2*gprsize]
mov src3q, [srcq+3*gprsize]
mov src4q, [srcq+4*gprsize]
mov src5q, [srcq+5*gprsize]
mov srcq, [srcq]
sub src1q, srcq
sub src2q, srcq
sub src3q, srcq
sub src4q, srcq
sub src5q, srcq
.loop:
cvtps2pi mm0, [srcq]
cvtps2pi mm1, [srcq+src1q]
cvtps2pi mm2, [srcq+src2q]
cvtps2pi mm3, [srcq+src3q]
cvtps2pi mm4, [srcq+src4q]
cvtps2pi mm5, [srcq+src5q]
packssdw mm0, mm3
packssdw mm1, mm4
packssdw mm2, mm5
pswapd mm3, mm0
punpcklwd mm0, mm1
punpckhwd mm1, mm2
punpcklwd mm2, mm3
pswapd mm3, mm0
punpckldq mm0, mm2
punpckhdq mm2, mm1
punpckldq mm1, mm3
movq [dstq ], mm0
movq [dstq+16], mm2
movq [dstq+ 8], mm1
add srcq, 8
add dstq, 24
sub lend, 2
jg .loop
emms
RET
%endmacro ; FLOAT_TO_INT16_INTERLEAVE6
%define pswapd PSWAPD_SSE
FLOAT_TO_INT16_INTERLEAVE6 sse
%define cvtps2pi pf2id
%define pswapd PSWAPD_3DN1
FLOAT_TO_INT16_INTERLEAVE6 3dnow
%undef pswapd
FLOAT_TO_INT16_INTERLEAVE6 3dn2
%undef cvtps2pi
%macro SCALARPRODUCT 1
; void add_int16(int16_t * v1, int16_t * v2, int order)
cglobal add_int16_%1, 3,3,2, v1, v2, order
shl orderq, 1
add v1q, orderq
add v2q, orderq
neg orderq
.loop:
movu m0, [v2q + orderq]
movu m1, [v2q + orderq + mmsize]
paddw m0, [v1q + orderq]
paddw m1, [v1q + orderq + mmsize]
mova [v1q + orderq], m0
mova [v1q + orderq + mmsize], m1
add orderq, mmsize*2
jl .loop
REP_RET
; void sub_int16(int16_t * v1, int16_t * v2, int order)
cglobal sub_int16_%1, 3,3,4, v1, v2, order
shl orderq, 1
add v1q, orderq
add v2q, orderq
neg orderq
.loop:
movu m2, [v2q + orderq]
movu m3, [v2q + orderq + mmsize]
mova m0, [v1q + orderq]
mova m1, [v1q + orderq + mmsize]
psubw m0, m2
psubw m1, m3
mova [v1q + orderq], m0
mova [v1q + orderq + mmsize], m1
add orderq, mmsize*2
jl .loop
REP_RET
; int scalarproduct_int16_sse2(int16_t * v1, int16_t * v2, int order, int shift)
cglobal scalarproduct_int16_%1, 3,3,4, v1, v2, order, shift
shl orderq, 1
add v1q, orderq
add v2q, orderq
neg orderq
movd m3, shiftm
pxor m2, m2
.loop:
movu m0, [v1q + orderq]
movu m1, [v1q + orderq + mmsize]
pmaddwd m0, [v2q + orderq]
pmaddwd m1, [v2q + orderq + mmsize]
paddd m2, m0
paddd m2, m1
add orderq, mmsize*2
jl .loop
%if mmsize == 16
movhlps m0, m2
paddd m2, m0
psrad m2, m3
pshuflw m0, m2, 0x4e
%else
psrad m2, m3
pshufw m0, m2, 0x4e
%endif
paddd m2, m0
movd eax, m2
RET
%endmacro
INIT_MMX
SCALARPRODUCT mmx2
INIT_XMM
SCALARPRODUCT sse2
; void ff_add_hfyu_median_prediction_mmx2(uint8_t *dst, const uint8_t *top, const uint8_t *diff, int w, int *left, int *left_top)
cglobal add_hfyu_median_prediction_mmx2, 6,6,0, dst, top, diff, w, left, left_top
movq mm0, [topq]
movq mm2, mm0
movd mm4, [left_topq]
psllq mm2, 8
movq mm1, mm0
por mm4, mm2
movd mm3, [leftq]
psubb mm0, mm4 ; t-tl
add dstq, wq
add topq, wq
add diffq, wq
neg wq
jmp .skip
.loop:
movq mm4, [topq+wq]
movq mm0, mm4
psllq mm4, 8
por mm4, mm1
movq mm1, mm0 ; t
psubb mm0, mm4 ; t-tl
.skip:
movq mm2, [diffq+wq]
%assign i 0
%rep 8
movq mm4, mm0
paddb mm4, mm3 ; t-tl+l
movq mm5, mm3
pmaxub mm3, mm1
pminub mm5, mm1
pminub mm3, mm4
pmaxub mm3, mm5 ; median
paddb mm3, mm2 ; +residual
%if i==0
movq mm7, mm3
psllq mm7, 56
%else
movq mm6, mm3
psrlq mm7, 8
psllq mm6, 56
por mm7, mm6
%endif
%if i<7
psrlq mm0, 8
psrlq mm1, 8
psrlq mm2, 8
%endif
%assign i i+1
%endrep
movq [dstq+wq], mm7
add wq, 8
jl .loop
movzx r2d, byte [dstq-1]
mov [leftq], r2d
movzx r2d, byte [topq-1]
mov [left_topq], r2d
RET
%macro ADD_HFYU_LEFT_LOOP 1 ; %1 = is_aligned
add srcq, wq
add dstq, wq
neg wq
%%.loop:
mova m1, [srcq+wq]
mova m2, m1
psllw m1, 8
paddb m1, m2
mova m2, m1
pshufb m1, m3
paddb m1, m2
pshufb m0, m5
mova m2, m1
pshufb m1, m4
paddb m1, m2
%if mmsize == 16
mova m2, m1
pshufb m1, m6
paddb m1, m2
%endif
paddb m0, m1
%if %1
mova [dstq+wq], m0
%else
movq [dstq+wq], m0
movhps [dstq+wq+8], m0
%endif
add wq, mmsize
jl %%.loop
mov eax, mmsize-1
sub eax, wd
movd m1, eax
pshufb m0, m1
movd eax, m0
RET
%endmacro
; int ff_add_hfyu_left_prediction(uint8_t *dst, const uint8_t *src, int w, int left)
INIT_MMX
cglobal add_hfyu_left_prediction_ssse3, 3,3,7, dst, src, w, left
.skip_prologue:
mova m5, [pb_7 GLOBAL]
mova m4, [pb_zzzz3333zzzzbbbb GLOBAL]
mova m3, [pb_zz11zz55zz99zzdd GLOBAL]
movd m0, leftm
psllq m0, 56
ADD_HFYU_LEFT_LOOP 1
INIT_XMM
cglobal add_hfyu_left_prediction_sse4, 3,3,7, dst, src, w, left
mova m5, [pb_f GLOBAL]
mova m6, [pb_zzzzzzzz77777777 GLOBAL]
mova m4, [pb_zzzz3333zzzzbbbb GLOBAL]
mova m3, [pb_zz11zz55zz99zzdd GLOBAL]
movd m0, leftm
pslldq m0, 15
test srcq, 15
jnz add_hfyu_left_prediction_ssse3.skip_prologue
test dstq, 15
jnz .unaligned
ADD_HFYU_LEFT_LOOP 1
.unaligned:
ADD_HFYU_LEFT_LOOP 0
|
